Quantitative Evaluation of Histomorphometry
3D Synchrotron-based Micro-tomography vs. 2D Histomorphometry
- Fig.1: Aerial view of the synchrotron light source BESSY with sketch of the X-ray beam path and photo of the SµCT stage at the BAMline .
- Fig.2: Volume rendering of a SµCT data set (7 µm pixel size) showing ex vivo bioceramic particles in a regenerating sheep mandible.
- Fig. 3: Left: histological image of a bone biopsy, the marked region was scanned with SµCT before sectioning; right: semi-automatic 2D image analysis in the marked region.
- Fig. 4: Left: the 3D rendered SµCT volume image of the bone biopsy, cropped along the slice which corresponds to the histological section (cf. fig.3); right: the 2D slice from the SµCT data set which corresponds to the histological image (cf. fig.3).
- The Bessy synchrotron light source in Berlin, Germany. Source: Bessy GmbH, Berlin
We use the intense light coming from a third generation synchrotron source to perform computed tomography (CT) with micro-resolution and high material contrast. This unique tool allows for shedding light on methodical errors which are involved when applying classical histomorphometry. The tomographic approach using X-rays is not only non-destructive, it also allows for an analysis in a true 3D manner. Thus, it delivers knowledge of which is important to fully understand e. g. bone regeneration.
Histomorphometry and hard tissue histology are established techniques for studying e.g. the local biodegradation of a bone substitute material . First studies comparing results achieved using 2D histomorphometry and 3D micro-CT already indicated that the sectioning of a specimen has a significant influence on the derived results [2, 3]. We use high-resolution, high-contrast synchrotron-based micro-CT (SµCT) and subsequent image analysis to compare, in a quantitative manner, histological results with a non-destructive, 3D method . Our analysis gives an approximate value of the error introduced when restricting investigations to 2D histology. As example we chose the study of bone regeneration in a defect after sinus floor elevation .
Synchrotron-based Micro-tomography and 3D Image Analysis
The use of synchrotron light with its high photon flux density, nearly parallel beam propagation and partial coherence allows for performing hard X-ray based imaging with spatial resolution up to 1 µm and different contrast modes [5, 6, 7, 8]. The high flux makes it feasible to use monochromators, hence reducing artefacts while increasing the image contrast. The influence of the finite X-ray source size can be suppressed by extending the distance between source and experiment to more than 100 m. The partial coherence gives access to more sensitive contrast modes like phase contrast . Our experiments were carried out at the BAMline of the Berlin Electron Storage Ring Company for Synchrotron Radiation (BESSY), Germany - see figure 1 . We scanned an embedded bone biopsy specimen with a monochromatic X-ray beam of 20 keV photon energy (absorption contrast) using an indirect pixel detector (1.5 µm pixel size, ~ 4 µm spatial detector resolution) [5, 9].
A volume of ca. 2.5 mm in diameter and 2.1 mm in height was imaged, defined by the finite beam height and the detector's field of view. The achieved data set was processed and analyzed using techniques derived from stochastic geometry: morphological structures were separated into Boolean images by a threshold hysteresis in combination with a region growing algorithm (voxels identified, e. g., as bone are marked with "1", remaining voxels are set to "0"). Further morphological transformations were applied to reduce noise [8, 10]. The 3D renderings were done using VolumeGraphics VGStudioMAX. An example image as derived by SµCT is shown in figure 2.
The specimen we analyzed is part of a comparative study where different biocompatible ceramics were applied after sinus floor elevation to locally support the bone regeneration [1, 4]. The bone biopsy was sampled 6 months after the operation using a trephine burr (size was approx. 2.5 mm in diameter and 5 mm in height). The tissue sample was immediately fixed in an ethanol-based fixative at room temperature for 5 days. This was followed by dehydratation and infiltration. Next the specimen was embedded in a resin. Then it was polymerized in polyethylene vials at 4 °C for 47 days. After polymerization the block was removed from the vials and excess resin was trimmed away. After completion of the SµCT scans the resin block which contained the biopsy was glued to acrylic slides using an epoxy resin based two-component adhesive. 50-µm-sections were cut using a Leitz 1600 sawing microtome. These sections were then grounded and polished with 1200 and 4000 grit silicone carbide paper. Prior to immunohistochemical staining, deacrylation of sections was performed by immersion in toluene, xylene and acetone. Subsequently they were rinsed in distilled water and placed in Tris-buffer (pH 7.4). Immunohistochemical staining was performed and Mayer's haematoxylin was used as a counterstain. The sections were measured semi-automatically using a light microscope (Olympus, Hamburg, Germany) in combination with a digital camera (Colourview III) and SIS AnalysisTM software (Olympus, Hamburg, Germany).
The imaging procedures result in 2D histological pictures and a tomographic volume data set. A typical histological image is displayed in figure 3 (left). A semi-automatic 2D analysis of the sample region which was scanned with SµCT as well (marked region in figure 3 - left) reveals a bone area fraction of 53.3 %. The morphological information of the bone structure in the SµCT image was then separated into a Boolean image. By manual scanning the tomographic slice corresponding to the histological image was identified (see figure 4). Plain counting of voxels in this slice delivered a bone area fraction of 54.9 %. We derive as a first result that 2D histomorphometry and the 2D analysis of the corresponding SµCT slice yield almost identical results. Having proved the compatibility of our SµCT approach in 2D, we can extend now to a true 3D approach, thus progressing to histovolumetry, as displayed in figure 4, by determining the bone volume fraction in the whole SµCT image: 48.8 % - a significantly smaller value.
Our findings strongly emphasize the demand for a 3D analysis to evaluate results achieved by classical histology. The determination of the bone fraction in a biopsy as performed with histology and SµCT showed that the extension from the 2D approach to 3D can change the value easily by approx. 10 %. While SµCT can not be used for every investigation due to the limited beam time available, it is a valuable tool which necessarily should be used to evaluate findings as derived by classical histology.
We like to acknowledge S. Zabler, J. Goebbels, L. Helfen, H. Riesemeier, B. R. Müller, Ch. Koch, A. Kopp, I. Borchert, and H. Benz.
 Knabe C., et al.: Biomat. 29, 2249-2258 (2008)
 Bonse U., et al.: Bone Miner. 25, 25-38 (1994).
 Müller R., et al., BONE 23, 59-66 (1998)
 Stiller M., et al., BONE 44, 619-628 (2009)
 Bonse U., Busch F., Prog. Biophys. Molec. Biol. 65, 133-169 (1996)
 Flannery B.P., et al.: Science 237, 1439-1444 (1987)
 Koch A., et al.: J. Opt. Soc. Am. 15, 19401-951 (1998)
 Banhart J., (ed.), Oxford University Press (2008)
 Rack A., et al.: Nucl. Inst. & Meth. in Res. A 586, 327-344 (2008)
 Schladitz K., Ohser J., Wiley-VCH, Weinheim (2009)